[Libre-soc-dev] setvl lost its reserved fields (was: setvl gaining CTR mode)

lkcl luke.leighton at gmail.com
Sun Jul 3 12:04:02 BST 2022

i know.  Jacob (sorry to be talking about you in the 3rd person, i'm aware you'll be reading this) has no idea of how much work is involved, and due to Asperger's he has consistently demonstrated that he is not able to take full context into consideration. it takes multiple repetitions to get through.

i very specifically added the CTR mode so that it did not make any changes at all to anything that you are doing, Dmitry (6 revisions so far, i believe: i don't really want a 7th without some feedback from Alan, Peter, or others)  the conditions are "vs=1,RA=0,RT!=0" which has nothing to do with sv_analysis, sv_binutils, no changes in fields, bit-encodings, nothing.  fields.txt does not change.  CSVs do not change.

even Jacob's idea to use one bit of SVi would have serious knock-on implications in binutils, unit tests, ISACaller, and TestIssuer, because that "one bit" has to be excluded from being allowed to be used.  syntax errors raised at the bare minimum in binutils.

basically Jacob, sorry to have to point this out, your working knowledge of SV is almost (or, over) a year out-of-date. with so many deadlines coming up (and legally-binding contractual committments made) i have to take the drastic but absolutely essential action of disregarding your input on the topic that you raised.

perhaps it would be better if you asked questions about why things are the way that they are. this would help clarify, help other people to understand as well, and bring you up to speed on almost a year's missing context.  that in turn would help clarify the spec.


On July 3, 2022 11:39:09 AM GMT+01:00, Dmitry Selyutin <ghostmansd at gmail.com> wrote:
>Before doing anything with SVi, please recall there are SVxd, SVxy,
>SVd. And, after all, let's keep it consistent. Please keep in mind that
>there are not only written specs, but also docs, tests and 3rd-party
>(e.g. binutils). That's not to even mention the need to reach an
>between all the parties involved.
>Best regards,

More information about the Libre-soc-dev mailing list